Merge pull request #1036 from yasoob/master

Added an IE and test for Criterion videos (closes #1035).

diff --git a/youtube_dl/extractor/__init__.py b/youtube_dl/extractor/__init__.py
index c00d5a3529a5c89c4e338c778ea9500fe9b936c7..494b1b9d390b3602ce03167d3a0e93224a94cd3e 100644 (file)
--- a/youtube_dl/extractor/__init__.py
+++ b/youtube_dl/extractor/__init__.py
@@ -8,6 +8,7 @@ from .breakcom import BreakIE
 from .brightcove import BrightcoveIE
 from .collegehumor import CollegeHumorIE
 from .comedycentral import ComedyCentralIE
+from .criterion import CriterionIE
 from .cspan import CSpanIE
 from .dailymotion import DailymotionIE
 from .depositfiles import DepositFilesIE

diff --git a/youtube_dl/extractor/criterion.py b/youtube_dl/extractor/criterion.py
new file mode 100644 (file)
index 0000000..a149d29
--- /dev/null
+++ b/youtube_dl/extractor/criterion.py
@@ -0,0 +1,40 @@
+# -*- coding: utf-8 -*-
+
+import re
+
+from .common import InfoExtractor
+
+class CriterionIE(InfoExtractor):
+    _VALID_URL = r'http://www.criterion.com/films/(.*)'
+    _TEST = {
+        u'url': u'http://www.criterion.com/films/184-le-samourai',
+        u'file': u'184.mp4',
+        u'md5': u'bc51beba55685509883a9a7830919ec3',
+        u'info_dict': {
+            u"title": u"Le Samouraï",
+            u"description" : u"In a career-defining performance, Alain Delon plays a contract killer with samurai instincts. A razor-sharp cocktail of 1940s American gangster cinema and 1960s French pop culture, maverick director Jean-Pierre Melville's masterpiece _Le Samouraï_ defines cool. "
+        }
+    }
+
+    def _real_extract(self, url):
+        mobj = re.match(self._VALID_URL, url)
+        video_id = mobj.group(1).split('-')[0]
+        webpage = self._download_webpage(url, video_id)
+
+        final_url = self._search_regex(r'so.addVariable\("videoURL", "(.+?)"\)\;',
+                                webpage, 'video url')
+        title = self._search_regex(r'<meta content="(.+?)" property="og:title" />',
+                                webpage, 'video title')
+        description = self._search_regex(r'<meta name="description" content="(.+?)" />',
+                                webpage, 'video description')
+        thumbnail = self._search_regex(r'so.addVariable\("thumbnailURL", "(.+?)"\)\;',
+                                webpage, 'thumbnail url')
+        ext = final_url.split('.')[-1]
+
+        return {'id': video_id,
+                'url' : final_url,
+                'title': title,
+                'ext': ext,
+                'description': description,
+                'thumbnail': thumbnail,
+                }